×

Skin Cancer Classification and Comparision from HAM10000 Dataset Images Using Ensemble of Convolutional Neural Networks

Author : Yogendra Sharma, Tushar Manger, Amar Sanyasi, Janiela Bhutia, Anushka Pradhan, Smriti Koirala Journa Name: INTERNATIONAL JOURNAL OF SCIENTIFIC RESEARCH AND ENGINEERING TRENDS Country : India Volume: 10 issue: 3 Year: 2024 Views : 583
Abstract:
Skin cancer is a growing global health concern, and early and accurate diagnosis is crucial for effective treatment. Convolutional Neural Networks (CNNs) have emerged as powerful tools for skin lesion classification. They offer the potential to improve diagnostic accuracy and assist dermatologists. This study com- pares the performance of five CNN architectures – DenseNet121, DEnseNet201, InceptionResNetV2, Xception, and SCC-NET on the preprocessed Ham10000 dataset. This dataset contains 10,000 dermoscopic images, categorized into seven skin lesion types, with 500 images per class. The objective is to identify the model that achieves the best accuracy and generalizability for this specific dataset. To evaluate the models, we use metrics like accuracy,, and F1-score. This research contributes to the growing body of knowledge on utilizing CNNs for skin cancer classification. It has the potential to pave the way for develop- ing reliable computer-aided diagnosis systems, which can further improve the accuracy of skin cancer diagnosis.
Leave Your Comment

Related Reviewers

Chat with Expert